OBJECTIVES: To use existing evidence on the drug-eluting stent (DES) CYPHER to assess its cost-effectiveness from a UK health service perspective. To analyse the degree of variation in cost-effectiveness by patient sub-group as represented by evidence from the alternative trials. METHODS: A probabilistic decision model was developed within which to synthesise available clinical, cost and utility evidence. The model compared the use of CYPHER and bare metal stents in the alternative patient sub-groups. Clinical data on rates of further revascularisation and non-fatal myocardial infarction were taken from three published randomised trials: RAVEL (largely patients with narrow vessels), and E-SIRIUS and SIRIUS (largely patients with long lesions). Resource costs of the initial procedures and all subsequent events were taken from NHS Reference Costs; manufacturers' list prices were used for the devices. A utility decrement, based on EQ-5D data collected in an earlier stent trial, was included for further revascularisation based on the expected duration of symptoms prior to re-treatment. Cost-effectiveness is reported in terms of the incremental cost per additional QALY, and cost-effectiveness acceptability curves are reported showing the probability of DES being the more cost-effective at particular threshold values for an additional QALY. RESULTS: Based on clinical data from RAVEL the cost per QALY was £30,400. The corresponding results for SIRIUS and E-SIRIUS were £10,500 and £4,950, respectively. Sensitivity analysis showed that when the clinically similar mortality rates observed in the trials were incorporated into the model, they had a major impact on mean cost-effectiveness but increased uncertainty. CONCLUSIONS: The incremental cost per QALY of DES, relative to bare metal stent, was consistently below £35,000 for all sub-groups. There was considerable uncertainty surrounding cost-effectiveness suggesting additional research may be appropriate.